The cost and limited availability of blood products and the small but finite risks associated with their administration demands that physicians use this vital resource prudently. Yet, little formal attention is given to appropriate blood utilization strategies during either medical school or post-graduate training. Consequently, transfusion strategies may be idiosyncratic, inconsistent and not well-grounded in scientific evidence. The underlying goal of this research is to improve transfusion practice by the application of knowledge-based techniques (ie, expert systems) to support the independent Teaming and decision making of physicians while actively engaged in clinical practice. The first specific aim is to implement a knowledge-based system for the evaluation of requests for platelet transfusion. An existing expert system that reflects current scientific knowledge as well as the clinical heuristics necessary for appropriate practice will be expanded and enhanced. The second aim is to develop a physician's workstation incorporating the expert system so as to provide a prompt and reasoned critique of platelet transfusion plans at the time the physician formulates a plan. The effects of the workstation on physician learning and transfusion practice will be evaluated in a randomized controlled clinical trial using a two period crossover design. Physicians on the bone marrow transplant, adult oncology, and pediatric hematology/oncology services in a university hospital will be studied. The project 9 will demonstrate a novel and generalizable approach to improving the education and practice of physicians using platelet transfusions in a way that realistically deals with the complexity, variability and scope of clinical problems. Experience gained here will apply directly to enhancing knowledge and appropriate use of other blood products, laboratory tests, and many other health care resources.